Cross-validating administrative and survey datasets through microsimulation in Luxembourg.

Ths chapter assesses the reliability of tax-benefi microsimulation based on survey and administrative data, using Luxembourg as a case study. Th Luxembourg household panel (PSELL) is the basis of the Luxembourg EU-SILC dataset, and is also the basis for the main tax-benefi microsimulation model in Luxembourg. Th survey's relatively small size raises concerns regarding the reliability and robustness. Th newly launched Luxembourg Social Security Data Warehouse based upon administrative data relating to the year 2003 can now be used as a potential alternative data source for microsimulation purposes. Ths chapter conducts extensive cross-validation of both data sources to examine their robustness and reliability. Th survey database performs reasonably well in capturing the relevant characteristics of the resident population and allows analyses with respect to characteristics not found in the administrative database, and vice versa. Th shapes of the equivalised income distributions broadly coincide but, we do observe a few important discrepancies at the extremes of the curves. Finally, through use of the EUROMOD microsimulation platform, we are able to show that the discrepancies observed between these income data sources are insuffient to signifiantly affct the conclusions drawn from analysis of policy alternatives.